Fig. 2.1 Communication setting of primary Bourne Channel (BC) in which the large network coverage and the wide communication range require the implementation of a large number of intermediate nodes that motivate a unified communication of control system components within a hybrid network.
2.2 Motivation: A hybrid network for NCSs
2.2.1 Communication constraints in NCSs
A study of networking aspects is always necessary for the design of wide-area NCSs. With regards to the needs of data exchange, each controller designed for a distributed NCS may have three basic actions [21]: (1) gather the information recorded by sensors; (2) interact with other controllers for coordination and cooperation; and (3) send control signals to the actuators. To achieve these actions, controllers commonly make use of efficient transmission techniques for point-to-point communication, sometimes over a long range and under challenging limitations on data rate, operational energy consumption, etc. [21, 131]. To give an example, some communication constraints are encountered in large-scale BC control system such aslong communication range and large network coverage,wireless-imposed network type,heterogeneous and resource-limited devices, dynamic topology, and lack of regular maintenance. The serious difficulties in the current state of the BC control system are frustrating some technological proposals in terms of interoperability, scalability, network design, and operations. Particularly for network routing, some standard protocols used in Internet stack (e.g., Routing Information Protocol - RIP, Open Shortest Path First - OSPF) or the protocols, especially designed for WSANs (e.g., Adhoc On-Demand Distance Vector - AODV, Dynamic MANET On-demand routing protocol - DYMO, Dynamic Source Routing - DSR) perform much worse for this task as discussed in [124] and some routing protocols are compared on the basis of different criteria in [55]. Dealing with these challenges requires the network design to adopt constraint-based networking technologies (such as LoRa® technology from Semtech providing long-range communications) or to optimize the dynamic routing process.
2.2.2 A hybrid network for heterogeneous components of the NCS
An NCS with communication constraints described above will take advantage of the unified communication among components within a hybrid network. This type of network refers to any network that contains two or more different communications standards or multiple topological structures [62, 154, 132]. It enables different types of subnetworks to be interconnected so that the components can share, coordinate and cooperate efficiently to accomplish different tasks. A hybrid network for NCS (as shown in Fig. 2.2) is commonly composed of traditional WSNs [92, 154], embedded networks (as defined in [122], Section 1.1) and Internet Protocol (IP) networks (e.g., local network infrastructure connected to the Internet) [132]. Although each subnetwork is
Fig. 2.2 An example of a hybrid network for NCS. It can be composed of traditional WSNs, embedded networks, and Internet Protocol (IP) networks. Intermediate nodes can route any data packets.
different in nature, but the network services are required to be homologous through the hybrid network.
To benefit from available technologies and the large-scale deployment of the Internet, it may be expected that NCS components should be integrated into Internet-connected networks [124].
In this integration, each component playing the role of a node in the global network, is able to dynamically join the Internet. In addition, with recent technological progress materializing IoT, the future Internet is foreseen to be extended as a worldwide network of interconnected devices or objects (as mentioned in [124]). Wireless network technology continues to play an ever-increasing role in data network architecture throughout the large range of electronic component markets. However, deploying an NCS configured to access the Internet (see Fig. 2.3) raises new challenges, which need to be tackled before such an integration. [124] highlighted and discussed the challenges newly emerging on the capabilities of WSAN devices such as the mobility management, quality of service (QoS), and network configuration in addition to their usual sensing/actuating functionality. Many open challenges remain, mostly due to the complex
2.2 Motivation: A hybrid network for NCSs | 39
Fig. 2.3 A hybrid network architecture composed of multiple Simple LoWPANs is proposed for NCS due to low mobility between subnetworks [122].
deployment characteristics and the stringent requirements imposed by various desired services of such hybrid network. For example, the control applications cannot directly profit by IP technology because of limited processing and transmission capabilities of nodes, non-permanent power source, small packet size or the overhead of IP header, etc. The promising 6LoWPAN technology allows IP networking for resource-limited devices.
2.2.3 6LoWPAN technology suitable for the NCS hybrid network
As surveyed in [122], some possible technologies can be considered while addressing the Internet integration of WSANs such as ZigBee (ZigBee Alliance), Machine-to-Machine (ETSI M2M), Future Internet (e.g., EU 4WARD project) and 6loWPAN (IETF). Nevertheless, non standard or especially non-IP networks tailored to specific applications seem inappropriate for building a global infrastructure. If all of the system components are connected through a single open standard protocol (e.g., IP), they might benefit without any extra effort, by seamless connectivity, unique addressability, and rich applicability. This shows a strong trend of convergence on standardization, both in industry and in academia, that is clearly steering towards an Internet-based approach. The 6LoWPAN and ROLL WGs of IETF are actively performing this standardization by specifying the 6LoWPAN technology and RPL protocol.
The design of a hybrid network for NCSs can take advantage of 6LoWPAN compared to other technologies (e.g., WiFi, ZigBee) such as scalability, self-organization, and low-power consumption [55, 122]. More advantages are presented in [122], Section 1.1.1. The 6LoW- PAN is expected to become a standard networking technology supported on any constrained network [122]. By this motivation, it shows up a wide range of interesting research proposals (e.g., RFC4919, RFC6568, RFC6606) and applications using 6LoWPAN (as surveyed in [55], and [122], Section 1.1.4). They offer solutions for making networks more reliable, adaptable, durable and manageable.
2.2.4 Different QoS requirements of control applications
A hybrid network is technologically realizable. However, one of the important concerns for the hybrid network design is the selection of network performance criteria. Whereas “QoS is the description and/or measurement of the performance of a network service as seen by network users” [51, 132], selecting network performance criteria for the hybrid network can be based on QoS requirements of control applications. For instance, the level meters or pressure meters in mesh-pressurized subnetworks requireminimum power consumption and minimum aggregation delay to reach their controllers. The control signals from controllers are expected to reach actuators (e.g., gate control motors) with theminimum packet loss rate, minimum aggregation delay, and in certain cases (e.g., with pressure regulating valves),minimum power consumption.
Whereas the controllers are implemented on devices usually attached to a power source, they only requireminimum aggregation delay and packet loss rate.
Eventually, the QoS-based performance criteria considered in the NCS hybrid network are packet loss rate,network latency, andnetwork lifetime. The challenging factors influencing on network performance in the hybrid network are the varying channel characteristics, bandwidth allocation, fault tolerance levels and hand-off support for heterogeneous wireless networks.
Network software may help to address those challenges by using dynamic routing.